A large variety of instruments and sensing principles are used in Earth observation. The key instrument features introduced here are:

Transcription

1 SECTION: Table_of_Contents_Chapter Chapter title in running head: CHAPTER 3. REMOTE-SENSING INSTRUMENTS Chapter_ID: 8_III_3_en Part title in running head: PART III. SPACE-BASED OBSERVATIONS SECTION: Chapter_book Chapter title in running head: CHAPTER 3. REMOTE-SENSING INSTRUMENTS Chapter_ID: 8_III_3_en Part title in running head: PART III. SPACE-BASED OBSERVATIONS CHAPTER 3. REMOTE-SENSING INSTRUMENTS This chapter provides an overview of basic instrument concepts. It introduces the high-level technical features of representative types of Earth observation instruments. 3.1 INSTRUMENT BASIC CHARACTERISTICS A large variety of instruments and sensing principles are used in Earth observation. The key instrument features introduced here are: (a) Scanning, swath and observing cycle; (b) Spectral range and resolution; (c) Spatial resolution (IFOV, pixel, MTF); (d) Radiometric resolution Scanning, swath and observing cycle The most characteristic feature of an instrument is the way in which it scans the scene to acquire the necessary observations. That depends on the type of orbit (low Earth orbit or geostationary Earth orbit), on the platform attitude control (spinning, three-axis stabilized), and sometimes on the specific type of measurement being taken. Only the most common scanning techniques will be described in this chapter. A driving requirement for scanning is whether the scene should be observed with continuity (imagery) or can be sampled (sounding). A similar scanning mode may be used in both cases. However, imagery requires continuous scene coverage, while sounding can accommodate sampling with gaps. In low Earth orbit (LEO) spacecraft, a 2D Earth scene scanning can use satellite motion for an along-track dimension. A cross-track scan can then be provided by a rotating scan mirror (Figure 3.1). Rotation speed is synchronized with satellite motion so that the cross-track image lines come out contiguously. ELEMENT 1: Floating object (Automatic) ELEMENT 2: Picture inline fix size Element Image: 8_III_3-1_en.eps END ELEMENT Figure 3.1. Typical scheme of a scanning radiometer in LEO END ELEMENT In the case of geostationary Earth orbit (GEO), there is no satellite motion with respect to the Earth s surface. The two scanning dimensions must therefore be generated either by the instrument or by satellite rotation. In Figure 3.2, it is assumed that the satellite is spin-stabilized: the west east scan is provided by the satellite rotation, while the north south scan is obtained by

2 2 VOLUME IV. SPACE-BASED OBSERVATIONS a step motor. For a three-axis stabilized platform, the instrument must generate both west east and north south scans. ELEMENT 3: Floating object (Automatic) ELEMENT 4: Picture inline fix size Element Image: 8_III_3-2_en.eps END ELEMENT Figure 3.2. Schematic scanning from a spin-stabilized GEO END ELEMENT The advent of array detectors has led to additional scanning possibilities. With LEO, a linear array can be placed orthogonally to the satellite track, and can scan the scene without any mechanical movement (pushbroom scanning). Under another scheme, the linear array can be placed parallel to the track; cross-track mechanical scanning will then scan several lines in parallel (whiskbroom scanning). With GEO, whiskbroom scanning is now the rule. For instance, SEVIRI on Meteosat Second Generation scans three lines in parallel for the infrared (IR) channels, and nine lines for the high-resolution visible spectrum (VIS) channel. A very convenient scanning mechanism for polarization-sensitive measurements is conical scanning (Figure 3.3). In this geometry, the incidence angle is constant. Therefore the effect of polarization does not change along the scan line (an arc). By contrast, the incidence angle for cross-track scanning changes along the scan line, when moving from the nadir to the image edge. This invariance of the polarization effect across the image is very important for microwave (MW) measurement in window channels, where radiation reflected from elements such as the sea surface is strongly polarized. The measurement of the differential polarization constitutes important information that would be very difficult to use if the incidence angle changed across the scene. Another interesting feature of conical scanning is that the resolution remains constant across the whole image. ELEMENT 5: Floating object (Top) ELEMENT 6: Picture inline fix size Element Image: 8_III_3-3_en.eps END ELEMENT Figure 3.3. Geometry of conical scanning END ELEMENT A disadvantage of conical scanning is that, with the selected incidence angle, the field of view does not normally reach the horizon. For example, the swath from an 800 km orbital height is ~1 600 km for a typical zenith angle of 53, which is optimal for enhancing differential polarization information in MW. By contrast, the swath for a cross-track scanning instrument is close to km, assuming a ± 70 zenith angle range, as shown in Volume IV, Chapter 2, Table 2.1. The swath is an important feature of an instrument in LEO since it determines the observing cycle. Volume IV, Chapter 2, Table 2.2 outlines that, for a Sun-synchronous orbit at 800 km, one instrument with a swath of at least km provides one global coverage per day for measurements operated in daytime only (for example, short-wave (SW) sensors), or two global coverages per day for measurements operated day and night (for example, IR or MW sensors). MW conical scanning instruments generally provide one global coverage per day. Estimating the observing cycle The order of magnitude of the observing cycle ( t, in days) for a given swath can be estimated by a simple calculation. Considering the Equator s length (~ km), the number of orbits per day (~14.2) and assuming that there is no significant overlap between adjacent swaths at the Equator, the calculation is as follows:

3 CHAPTER 3. REMOTE-SENSING INSTRUMENTS 3 (a) For day and night sensors (IR or MW) operating on both ascending and descending passes: t = 1 400/swath (for example, for a MW conical scanner with km swath: t = 1 day); (b) For daytime only sensors (SW) operating on only one pass per orbital period: t = 2 800/swath (for example, for VIS land observation with 180 km swath: t = 16 days). The concept of swath is not applicable for instruments with no cross-track scanning, such as altimeters or cloud radars. Where that is the case, the cross-track sampling interval x at the Equator replaces swath in the relationship above. It is also useful to estimate the global average of this sampling interval. The interval is given by a slightly different relationship because of shorter orbit spacing at higher latitudes: (a) At an average cross-track sampling interval x, the typical time needed for global coverage is: (i) t = 900/ x for day and night sensors (for example, the Environmental Satellite (Envisat) Radar Altimeter 2 (RA-2): x = 26 km, t = 35 days); (ii) t = 1 800/ x for daytime only (for example, the National Oceanic and Atmospheric Administration (NOAA) Solar Backscatter Ultraviolet (SBUV) instrument: x = 170 km, t = 11 days). (b) Reciprocally, in a time interval t (for example, the orbit repeat cycle or a main sub-cycle) the average cross-track sampling interval obtained is: (i) x = 900/ t for day and night sensors (for example, the Joint Altimetry Satellite Oceanography Network (JASON) altimeter: t = 10 days, x = 90 km); (ii) x = 1 800/ t for daytime only (for example, NOAA SBUV: t = 5 days, x = 360 km). Limb sounders are generally considered as non-scanning instruments in the cross-track direction. Assuming a cross-track sampling interval x = 300 km which is equal to the horizontal resolution of the measurements, the relationships above yield, taking MIPAS (which flew on the ENVISAT satellite) as an example, the following observing cycle: - t 3 days for day and night sensors (for example, MIPAS on Envisat); For instruments providing sparse but well-distributed observations, the coverage cycle or the average sampling can be estimated by comparing the number of events and their resolution with the Earth s surface to be covered. In the example of radio occultation using GPS and GLONASS, each satellite is able to provide about observations per day, with a typical measurement resolution of 300 km for a total Earth s surface of 510 million km 2. Therefore: (a) Time required with one satellite providing observations/day: t = /300/300/1 000 = 5.7 days; (b) Number of satellites required for an observing cycle t: N = 5.7/ t (for example, for t = 0.5 days, the number of satellites is close to 12). Sun-, moon- or star-occultation instruments are an extreme case. Sun or moon occultation provides very few measurements per day, and only at the high altitudes of the day/night terminator in the case of the Sun, or at somewhat lower latitudes for the moon. Star occultation may provide several tens of measurements per day (for example, 40 for the Envisat Global Ozone Monitoring by Occultation of Stars (GOMOS)), evenly distributed by latitude Spectral range: radiometers and spectrometers Another major characteristic feature of an instrument is the spectral range over which it operates. As discussed in Volume IV, Chapter 2, 2.2, the spectral range determines which of the body s

4 4 VOLUME IV. SPACE-BASED OBSERVATIONS properties can be observed including reflectance, temperature and dielectric properties. Within the spectral range, there may be window regions and absorption bands, which mainly address condensed or gaseous bodies, respectively. A spectral range may be more or less narrow, depending on the effects to be enhanced, or on the disrupting factors that need to be eliminated. The sub-divisions of a band or a window covered by an instrument are called channels. The number of channels depends on the number of pieces of independent information which have to be extracted from one band. If a limited number of wellseparated channels are sufficient for the purpose, the instrument may only include those channels, and it is called a radiometer. If the information content rapidly changes with the frequency along the spectral range to the extent that channels must be contiguous, the instrument is called a spectrometer. The technique adopted for channel separation, or for spectrometer subrange separation, is a major characteristic of the instrument. There are essentially two possible ways to physically separate two channels towards individual detectors (or detector arrays) and associated filter systems. First, the beam could be focused on a field stop and split into two bands by a dichroic mirror. The advantage is that co-registration is ensured, as the two channels look at the same field of view (which could comprise an array of instantaneous fields of view (IFOVs)). However, if the two wavelengths are too close to each other (for example, a split window), a dichroic mirror cannot separate them sharply enough. The second solution is to let the full beam produce the image in the focal plane and set detectors (or detector arrays) with different filters (thus identifying different channels) in different parts of the focal plane (in-field separation). It is a much simpler solution, but as each channel looks at a different IFOV, co-registration problems can occur. Combined solutions are possible: Figure 3.4 illustrates the solution implemented in the Meteosat Second Generation SEVIRI to separate the eight IR channels. Each channel is viewed by three detectors. Current detector arrays are much larger than they have been in the past, which makes in-field separation more convenient. ELEMENT 7: Floating object (Automatic) ELEMENT 8: Picture inline fix size Element Image: 8_III_3-4_en.eps END ELEMENT Figure 3.4. Channel separation scheme in SEVIRI on Meteosat Second Generation END ELEMENT Spectrometers provide continuous spectral sampling within the spectral range or in a number of spectral subranges. (These subranges are sometimes called channels and should not be confused with the channels of a radiometer). There are several types of spectrometer, the simplest of which is the prism. Others include grating spectrometers and interferometers; the most common interferometers are the Michelson and the Fabry-Pérot. Figure 3.5 shows the scheme of a Michelson interferometer. ELEMENT 9: Floating object (Automatic) ELEMENT 10: Picture inline fix size Element Image: 8_III_3-5_en.eps END ELEMENT Figure 3.5. Scheme of a Michelson interferometer emphasizing the two input and output ports END ELEMENT The spectral resolution of a spectrometer is an important feature. The resolution of a Michelson interferometer is determined by the maximum length of the optical path difference (OPD) between the rays reflected by the fixed and moving mirrors. Referring to the unapodized resolution, the formula is:

5 CHAPTER 3. REMOTE-SENSING INSTRUMENTS 5 1OPD max (3.1) For example, in IASI, an instrument mounted on the Meteorological Operational (Metop) satellite, the excursion of the moving mirror is ± 2 cm. Therefore, OPD max = 4 cm and ν = 0.25 cm 1. If fine analysis of the spectrum is needed (for instance, to detect trace gas lines), apodization that implies a factor ~2 is required. The apodized resolution is therefore ν = 0.5 cm 1. For a grating spectrometer, the resolution is determined by the number of grooves, N, and the chosen order of diffraction, m. The resolving power / is given by: m N (3.2) For a Michelson interferometer, the spectral resolution is constant with a variable wavelength. For a grating spectrometer, however, it is the resolving power that is constant, and the spectral resolution that changes with wavelength. If a grating spectrometer is to cover a wide spectral range, that range should be subdivided into subranges that use different orders of diffraction, m. Resolving power may thus change based on subrange. For a radiometer, the number of channels and their bandwidths play an equivalent role to spectral resolution for a spectrometer Spatial resolution (instantaneous field of view, pixel, and the modulation transfer function) Colloquially, spatial resolution means the association of different features. The IFOV is probably the closest to what is commonly meant by resolution. In optical instruments (i.e. SW and IR instruments), it is determined by the beamwidth of the optics and the size of the detector. In MW it is determined by the size of the antenna. In optical systems, the size of the IFOV is designed primarily on the basis of energy considerations (see 3.1.4). The IFOV may be determined by the shape of the detector; and although that may be square, the contour of an IFOV is not unduly sharp. In fact, the image of a point is a diffraction figure called point spread function (Figure 3.6). The IFOV is the convolution of the point spread function and the spatial response of the detector. ELEMENT 11: Floating object (Bottom) ELEMENT 12: Picture inline fix size Element Image: 8_III_3-6_en.eps END ELEMENT Figure 3.6. Shape of the point spread function END ELEMENT The energy entering the detector is also determined by the integration time between successive signal readings. During image scanning, the position of the line of sight will change by an amount called sampling distance. When plotted in a rasterized 2D pattern, a pixel (picture element) appears as a series of rectangular elements. In the x-direction, they correspond to the sampling distance and in the y-direction, to the satellite motion during the time distance from one line to the next (or the step motion in the north south direction from GEO). The pixel is often confused with the resolution. That is because users can directly perceive the size of the picture element, whereas the IFOV is an engineering parameter that the user cannot see. Nevertheless, it is wrong to think that resolution can be improved by reducing the integration time, since the integration time must be suitable to ensure an appropriate radiometric accuracy (see 3.1.4). There is a balance between the size of the IFOV and the size of the pixel. When using a perfect imager, sampling is performed so that the sampling distance is equal to the IFOV size. That

6 6 VOLUME IV. SPACE-BASED OBSERVATIONS means that the IFOVs are continuous and contiguous across the image. Otherwise the image may be oversampled (pixel < IFOV, i.e. there is overlap between successive IFOVs) or undersampled (pixel > IFOV, i.e. there are gaps between successive IFOVs). Oversampling is useful to reduce aliasing effects (undue enhancement of high spatial frequencies because of reflection from the border). Undersampling may be necessary when more energy must be collected in order to ensure the required radiometric accuracy. Examples of relationships between IFOV and pixel are: (a) AVHRR IFOV: 1.1 km; pixel: 1.1 km along-track, 0.80 km along-scan (oversampled); (b) SEVIRI IFOV: 4.8 km; pixel: 3.0 km across-scan and along-scan (oversampled). The modulation transfer function (MTF) is closely linked to the concept of IFOV and pixel. It is even more directly linked to the size L of an instrument s primary optics. The MTF represents the capability of the instrument to correctly manage the response to amplitude variation at the scene. It is the ratio between the observed amplitude, and the true signal amplitude from the scene, figured as sinusoidal. The observed amplitude is damped due to factors such as the diffraction from the optical aperture, the window introduced by the detector and smearing from integrating electronics. The effect of integrating the radiation over the (squared) detector introduces a contribution to the MTF: sin y MTF window ( f) sinc( IFOV f), with f 1 2 x km and sinc y (3.3) y 1 This shows that, for x = IFOV, MTF = sinc ( /2) = 2/. Therefore, even for a perfect imager, MTF is lower than unity. The value 2/ 0.64 corresponds to the area of a half sinusoid inscribed in a square. The concept of MTF must be seen as closely associated with radiometric accuracy: it specifies at which spatial wavelength two features are actually resolved if their radiation differs by just the detectable minimum. Two features whose radiation differs by substantially more than the detectable minimum can be resolved, even if they are substantially smaller. However, if they are as small as IFOV/2, then MTF = 0, and they can in no way be resolved (f = 1/IFOV is called the cut-off frequency). It is interesting to note how MTF window changes at different spatial wavelengths measured in terms of IFOV (where spatial wavelength is 2 x). Table 3.1 can be derived from equation 3.3: Table 3.1. Variation of the MTF window function of the ratio x/ifov x/ifov 1/2 2/ MTF window This shows that features as small as two thirds of the IFOV can be resolved, but only if their radiances differ by more than three times the detectable minimum. It also shows that features twice as large as the IFOV can be resolved if their radiance difference exceeds the minimum by 10%. The other major contribution to the MTF is diffraction. The relation is: 2 1 f f f 2 MTFdiffraction cos ( ) 1 ( ) fd fd fd (3.4) where f = 1/(2 H) (with H = satellite height); = angular resolution (i.e. IFOV/H); f d = L/( H); L = aperture of the primary optics; and thus f/f d = /(2 L). Diffraction is dominant when the wavelength is relatively large (as with microwaves), when the optics aperture is relatively small, or when the satellite altitude is relatively large (as with GEO). The value MTF diffraction = 0.5 occurs for f/f d = 0.404, that is:

7 CHAPTER 3. REMOTE-SENSING INSTRUMENTS (3.5) L which is the classical law of diffraction. In summary, what is commonly meant by the term resolution involves at least three parameters. Although they should be considered in context, each one is more closely associated with a different perception: (a) IFOV: not visible to the user; controls the radiometric budget of the image; (b) Pixel: provides direct perception of the degree of detail in the image; (c) MTF: by monitoring the amplitude restitution, provides the perception of contrast Radiometric resolution Although scarcely visible to the user, radiometric resolution is a defining element of instrument design. Scanning mechanisms, spectral resolution, spatial resolution, integration time and optics apertures are all designed to fulfil radiometric resolution requirements. Radiometric resolution is the minimum radiance difference necessary to distinguish two objects in two adjacent IFOVs. The observed difference is a combination of the true radiance difference between two bodies (signal) and the difference observed even when the contents of the IFOVs are identical (noise). The signalto-noise ratio (SNR) is one way of expressing radiometric resolution. Noise is a function of several factors, as set out in the radiometric performance formula: NESR 2 F D* t (3.6) where: NESR = noise equivalent spectral radiance (unit: W m 2 sr 1 [cm 1 ] 1, i.e. per unit of wave number); F = f/l, F-number (f = system focal length, L = telescope aperture); D* = detectivity (strongly depending on ν); ν = spectral resolution (expressed in terms of wave number ν = 1/ ); = instrument transmissivity; t = integration time; = system throughput, given by the product of ( L 2 /4) by (IFOV 2 /H 2 ); where: L 2 /4 = areal aperture of the telescope; H = satellite altitude; IFOV 2 /H 2 = solid angle subtended by the IFOV. In general, when defining I(ν) = spectral radiance at instrument input (unit: W m 2 sr 1 [cm 1 ] 1 ), this leads to: SNR I NESR (3.7) For SW, the input radiance is the solar spectral radiance corrected for the incidence angle and reflected according to body reflectivity (or albedo, if the body can be approximated as a Lambertian diffuser). Equations 3.6 and 3.7 lead to: IFOV SNR t L (at a specific input radiance I or albedo ) (3.8) That relationship explicitly links user-oriented parameters such as SNR, spectral resolution ν, IFOV and integration time t to the size of the primary optics L.

8 8 VOLUME IV. SPACE-BASED OBSERVATIONS In the case of narrowband IR channels, radiometric resolution is usually quoted as: NESR NE T (3.9) db / dt where B = Planck function, and NE T = noise equivalent differential temperature at a specific temperature T. With NE T, the performance formula 3.6 can be rewritten as follows: 4H F NE T IFOV t db / dt L D* (3.10) The left-hand side of formula 3.10 shows user-oriented parameters (radiometric, spectral, horizontal and time resolutions), while the right-hand side shows instrument sizing parameters (F-number, optics aperture, detectivity and transmissivity). This formula is not valid in all circumstances, but is instructive for a rough analysis in many instances. Cases where the formula is not valid mainly occur when the detector itself constitutes the dominant noise source, or when the detector response time is not short enough in comparison with the available integration time. This is normally the case in the far infrared (FIR) range. But it may also be the case with shorter wavelengths if, for instance, microbolometers or thermal detectors are used in order to operate at room temperature. In other words, D*, which obviously depends on ν, may be heavily dependent on the available integration time. In the case of broadband channels, the concept of NESR, as expressed in equation 3.6, must be redefined in terms of integrated noise over the full spectral range of each channel. In that case, it is also possible to obtain a relationship similar to formula 3.10: 1 NE R IFOV t (3.11) L where NE R = noise equivalent differential radiance (unit: W m 2 sr 1 ). The situation is different in the microwave range for two reasons. First, the need to limit the antenna size means that diffraction law establishes a link between the IFOV and the optics aperture L: IFOV 1.24 H c L * (3.12) where ν* = frequency = c/ ; and c = speed of light. Thus, there is less latitude for trade-off parameters. Second, the detection mechanism is based on comparing the scene temperature with the "system temperature", which increases with the bandwidth. The final outcome is that the equivalent of equations 3.8, 3.10 and 3.11 in the MW range is: where T sys = system temperature. NE T * t T sys (3.13) The system temperature depends on many technological factors, and increases sharply as frequency increases. On the one hand, equation 3.13 shows that, in the case of MW, radiometric resolution can only marginally be improved by increasing bandwidth and integration time, since the benefit only grows with the square root. On the other hand, because of the diffraction-limited regime, the usual way to increase SNR by increasing optics aperture is not applicable. That is because, if the antenna diameter is increased, the IFOV is reduced commensurately (see equation 3.12).

9 CHAPTER 3. REMOTE-SENSING INSTRUMENTS 9 This short review highlights the direct impact of user and mission requirements on instrument sizing. In addition, it shows how important it is to formulate requirements in a way that leaves room for optimization, without necessarily compromising the overall required performance. For instance with reference to equation 3.10, it is possible to draw a number of conclusions. (a) For a given set of instrument parameters (L, F, and D*), some user-driven parameters (NE T, ν, IFOV and t) can be enhanced at the expense of others. In certain cases, this can be done at the software level during data processing on the ground. However, if all user requirements become more demanding and no compromises are made, a larger instrument will be necessary. (b) The effect of NE T, ν and IFOV on instrument size is linear in relation to the optics diameter L. The effect of t (the integration time, driven by the requirement to cover a given area in a given time) is damped by the square root. Therefore, requirements for increased coverage and/or more frequent observation have a lesser impact than requirements for improved spatial, spectral and radiometric resolution. (c) Increasing the optics aperture L has a significant impact on instrument size. Since it is very difficult to implement optical systems with F-number = f/l < 1, an increase in L implies an increased focal length, and therefore a volumetric growth of overall instrument optics. For example, a reduction in the IFOV from three to two kilometres would double the instrument mass. 3.2 INSTRUMENT CLASSIFICATION In this section, Earth observation instruments are classified according to their main technical features. The following instrument types are considered: (a) Moderate-resolution optical imager (b) High-resolution optical imager (c) Cross-nadir scanning short-wave sounder (d) Cross-nadir scanning infrared sounder (e) Microwave-imaging radiometer or microwave-sounding radiometer (f) Limb sounder (g) Global navigation satellite system radio-occultation sounder (h) Broadband radiometer (i) (j) Solar irradiance monitor Lightning imager (k) Cloud radar and precipitation radar (l) Radar scatterometer (m) Radar altimeter (n) Imaging radar (synthetic aperture radar) (o) Space lidar

10 10 VOLUME IV. SPACE-BASED OBSERVATIONS (p) Gravity sensor (q) Solar activity, solar wind or deep space monitor (r) Space environment monitor (s) Magnetosphere or ionosphere sounder Most instrument types are subdivided into more detailed categories. Examples are provided to illustrate how instrumental features can be suited to particular applications. A comprehensive list of satellite Earth observation instruments, with their detailed descriptions, is available through the WMO online database of space-based capabilities, available from the WMO Space Programme website Moderate-resolution optical imager This instrument has the following main characteristics: (a) Operates in the VIS, near infrared (NIR), short-wave infrared (SWIR), medium-wave infrared (MWIR) and thermal infrared (TIR) bands (i.e. from 0.4 to 15 µm); (b) Discrete channels, from a few to several tens, separated by dichroics, filters or spectrometers, with bandwidths from ~10 nm to ~1 µm; (c) Imaging capability: continuous and contiguous sampling, with spatial resolution in the order of one kilometre, covering a swath of several hundred to a few thousand kilometres; (d) Scanning: generally cross-track, but sometimes multiangle, and sometimes under several polarizations; (e) Applicable in both LEO and GEO. Depending on the spectral bands, number and bandwidth of channels, and radiometric resolution, the application fields may include: (a) Multi-purpose VIS/IR imagery for cloud analysis, aerosol load, sea-surface temperature, seaice cover, land-surface radiative parameters, vegetation indexes, fires, and snow cover. The extent of the spectral range is a critical instrument feature; (b) Ocean-colour imagery, aerosol observation and vegetation classification. The number of channels with narrow bandwidth in VIS and NIR is a critical instrument feature; (c) Imagery with special viewing geometry, for the best observation of aerosol and cirrus, accurate sea-surface temperature, land-surface radiative parameters including bidirectional reflectance distribution function. The critical instrument features are the number of viewing angles and, when available, polarizations. Tables describe three examples of multi-purpose VIS/IR imagers (AVHRR/3 in LEO, MODIS in LEO and SEVIRI in GEO), one example of an ocean-colour imager (MERIS) and one example of an imager with special viewing geometry (POLDER). MODIS is an experimental sensor and plays a particular role as a wide scope multi-purpose VIS/IR imager. It is largely used to help define the specifications of follow-on operational instruments. The main uses of its various groups of channels are highlighted in Table 3.3. SECTION: Ignore_book Chapter title in running head: CHAPTER 3. REMOTE-SENSING INSTRUMENTS Chapter_ID: 8_III_3_en Part title in running head: PART III. SPACE-BASED OBSERVATIONS

11 CHAPTER 3. REMOTE-SENSING INSTRUMENTS 11 Table 3.2. Example of multi-purpose VIS/IR imager operating in LEO: AVHRR/3 on NOAA and Metop AVHRR/3 Advanced Very High Resolution Radiometer 3 Satellites Scanning technique NOAA-15, NOAA-16, NOAA-17, NOAA-18, NOAA-19, Metop-A, Metop-B, Metop-C Multi-purpose VIS/IR imagery for cloud analysis, aerosol load, sea-surface temperature, sea-ice cover, land-surface radiative parameters, normalized difference vegetation index, fires, snow cover, etc. 6 channels (channel 1.6 and 3.7 are alternative), balanced VIS, NIR, SWIR, MWIR and TIR Cross-track: pixels of 800 m SSP (sub-satellite point) with a swath of km Along-track: six 1.1 km lines a second Global coverage twice a day (LW (long-wave) channels) or once a day (SW channels) 1.1 km IFOV Mass: 33 kg Power: 27 W Data rate: kbps Central wavelength Spectral interval NEΔT or SNR at specified input spectral radiance µm µm 9 at 0.5 % albedo µm µm 9 at 0.5 % albedo 1.61 µm µm 20 at 0.5 % albedo 3.74 µm µm 0.12 K at 300 K µm µm 0.12 K at 300 K µm µm 0.12 K at 300 K Table 3.3. Example of multi-purpose VIS/IR imager operating in LEO: MODIS on EOS-Terra and EOS-Aqua MODIS Satellites Scanning technique Moderate-resolution Imaging Spectroradiometer EOS-Terra and EOS-Aqua Multi-purpose VIS/IR imagery for cloud analysis, aerosol properties, sea- and land-surface temperature, sea-ice cover, ocean colour, land-surface radiative parameters, vegetation indexes, fires, snow cover, total ozone, cloud motion winds in polar regions, etc. 36-channel VIS, NIR, SWIR, MWIR and TIR spectroradiometer Whiskbroom: a 19.7 km-wide strip of along-track is cross-track scanned every s. The strip includes 16 parallel lines sampled by pixels of SSP, or 32 parallel lines sampled by pixels of 500 m SSP, or 64 parallel lines sampled by pixels of 250 m SSP, with a swath of km. Global coverage nearly twice a day (LW channels) or once a day (SW channels)

12 12 VOLUME IV. SPACE-BASED OBSERVATIONS IFOV: 250 m (two channels), 500 m (5 channels), (29 channels) Mass: 229 kg Power: 225 W Data rate: 11 Mbps (daytime); 6.2 Mbps (average) SECTION: Ignore_book Chapter title in running head: CHAPTER 3. REMOTE-SENSING INSTRUMENTS Chapter_ID: 8_III_3_en Part title in running head: PART III. SPACE-BASED OBSERVATIONS Central wavelength Spectral interval NEΔT or SNR at specified input spectral radiance IFOV at SSP Primary use µm µm µm µm 128 at 21.8 W m 2 sr 1 µm m 201 at 24.7 W m 2 sr 1 µm m Land/Cloud/ Aerosol boundaries µm µm µm µm µm µm µm µm µm µm 243 at 35.3 W m 2 sr 1 µm at 29.0 W m 2 sr 1 µm 1 74 at 5.4 W m 2 sr 1 µm at 7.3 W m 2 sr 1 µm at 1.0 W m 2 sr 1 µm m 500 m 500 m 500 m 500 m Land/Cloud/ Aerosol properties µm µm µm µm µm µm µm µm µm µm µm µm µm µm µm µm µm µm 880 at 44.9 W m 2 sr 1 µm at 41.9 W m 2 sr 1 µm at 32.1 W m 2 sr 1 µm at 27.9 W m 2 sr 1 µm at 21.0 W m 2 sr 1 µm at 9.5 W m 2 sr 1 µm at 8.7 W m 2 sr 1 µm at 10.2 W m 2 sr 1 µm at 6.2 W m 2 sr 1 µm 1 Ocean colour, Phytoplankton, Biogeochemistry µm µm µm µm µm µm 167 at 10.0 W m 2 sr 1 µm 1 57 at 3.6 W m 2 sr 1 µm at 15.0 W m 2 sr 1 µm 1 Atmospheric water vapour 3.75 µm 3.96 µm 3.96 µm 4.06 µm µm µm µm µm 0.05 K at 0.45 W m 2 sr 1 µm K at 2.38 W m 2 sr 1 µm K at 0.67 W m 2 sr 1 µm K at 0.79 W m 2 sr 1 µm 1 Surface/Cloud temperature 4.47 µm 4.55 µm µm µm 0.25 K at 0.17 W m 2 sr 1 µm K at 0.59 W m 2 sr 1 µm 1 Atmospheric temperature µm 6.77 µm 7.33 µm µm µm µm 150 at 6.0 W m 2 sr 1 µm K at 1.16 W m 2 sr 1 µm K at 2.18 W m 2 sr 1 µm 1 Cirrus clouds, Water vapour 8.55 µm µm 0.25 K at 9.58 W m 2 sr 1 µm 1 Cloud properties 9.73 µm µm 0.25 K at 3.69 W m 2 sr 1 µm 1 Ozone µm µm µm µm 0.05 K at 9.55 W m 2 sr 1 µm K at 8.94 W m 2 sr 1 µm 1 Surface/Cloud temperature µm µm µm µm µm µm µm µm 0.25 K at 4.52 W m 2 sr 1 µm K at 3.76 W m 2 sr 1 µm K at 3.11 W m 2 sr 1 µm K at 2.08 W m 2 sr 1 µm 1 Cloud-top temperature SECTION: Ignore_book Chapter title in running head: CHAPTER 3. REMOTE-SENSING INSTRUMENTS

13 CHAPTER 3. REMOTE-SENSING INSTRUMENTS 13 Chapter_ID: 8_III_3_en Part title in running head: PART III. SPACE-BASED OBSERVATIONS Table 3.4. Example of multi-purpose VIS/IR imager operating in GEO: SEVIRI on Meteosat Second Generation SEVIRI Satellites Scanning technique Spinning Enhanced Visible Infrared Imager Meteosat-8, Meteosat-9, Meteosat-10, Meteosat-11 Multi-purpose VIS/IR imagery for cloud analysis, aerosol load, sea-surface temperature, land-surface radiative parameters, normalized difference vegetation index, fires, snow cover, wind from cloud motion tracking, etc. 12 channels, balanced VIS, NIR, SWIR, MWIR and TIR Mechanical Spinning satellite E W continuous S N stepping Full disk every 15 min Limited areas in correspondingly shorter time intervals 4.8 km IFOV, 3 km sampling for 11 narrow channels 1.6 km IFOV, 1 km sampling for 1 broad VIS channel Mass: 260 kg Power: 150 W Data rate: 3.26 Mbps Central wavelength Spectral interval (99% encircled energy) SNR or NEΔT at specified input radiance Not applicable (broad bandwidth channel) µm 4.3 at 1 % albedo µm µm 10.1 at 1 % albedo 0.81 µm µm 7.28 at 1 % albedo 1.64 µm µm 3 at 1 % albedo 3.92 µm µm 0.35 K at 300 K 6.25 µm µm 0.75 K at 250 K 7.35 µm µm 0.75 K at 250 K 8.70 µm µm 0.28 K at 300 K 9.66 µm µm 1.50 K at 255 K 10.8 µm µm 0.25 K at 300 K 12.0 µm µm 0.37 K at 300 K 13.4 µm µm 1.80 K at 270 K SECTION: Ignore_book Chapter title in running head: CHAPTER 3. REMOTE-SENSING INSTRUMENTS Chapter_ID: 8_III_3_en

14 14 VOLUME IV. SPACE-BASED OBSERVATIONS Part title in running head: PART III. SPACE-BASED OBSERVATIONS Table 3.5. Example of ocean-colour imager operating in LEO: MERIS on Envisat MERIS Satellite Scanning technique Medium Resolution Imaging Spectrometer Envisat Ocean-colour imagery, aerosol properties, vegetation indexes, etc. 15 very narrow-bandwidth VIS and NIR channels Pushbroom pixel/line (split into 5 parallel optical systems) Total swath: km Global coverage in 3 days in daylight Basic IFOV 300 m Reduced resolution for global data recording: m Mass: 200 kg Power: 175 W Data rate: 24 Mbps Central wavelength Bandwidth SNR at specified input spectral radiance nm 10 nm at 47.9 W m 2 sr 1 µm nm 10 nm at 41.9 W m 2 sr 1 µm nm 10 nm at 31.2 W m 2 sr 1 µm nm 10 nm at 23.7 W m 2 sr 1 µm nm 10 nm at 18.5 W m 2 sr 1 µm nm 10 nm 863 at 12.0 W m 2 sr 1 µm nm 10 nm 708 at 9.2 W m 2 sr 1 µm nm 7.5 nm 589 at 8.3 W m 2 sr 1 µm nm 10 nm 631 at 6.9 W m 2 sr 1 µm nm 7.5 nm 486 at 5.6 W m 2 sr 1 µm nm 3.75 nm 205 at 3.4 W m 2 sr 1 µm nm 15 nm 628 at 4.9 W m 2 sr 1 µm nm 20 nm 457at 3.2 W m 2 sr 1 µm nm 10 nm 271 at 3.1 W m 2 sr 1 µm nm 10 nm 211 at 2.4 W m 2 sr 1 µm 1 SECTION: Ignore_book Chapter title in running head: CHAPTER 3. REMOTE-SENSING INSTRUMENTS Chapter_ID: 8_III_3_en Part title in running head: PART III. SPACE-BASED OBSERVATIONS

15 CHAPTER 3. REMOTE-SENSING INSTRUMENTS 15 Table 3.6. Example of imager with special viewing geometry: POLDER on PARASOL POLDER Satellite Scanning technique Polarization and Directionality of the Earth s Reflectances Polarisation et anisotropie des réflectances au sommet de l'atmosphère, couplées avec un satellite d'observation emportant un lidar (PARASOL) Imagery with special viewing geometry, for best observation of aerosol and cirrus, land surface radiative parameters including bidirectional reflectance distribution function, etc. Bidirectional viewing Multipolarization 9 narrow-bandwidth VIS and NIR channels 242 x 274 charge-coupled device (CCD) arrays Swath: km Each Earth s spot viewed from more directions as satellite moves Near-global coverage every day in daylight 6.5 km IFOV Mass: 32 kg Power: 50 W Data rate: 883 kbps Central wavelength Bandwidth No. of polarizations SNR at specified input spectral radiance nm 13.4 nm 200 at 61.9 W m 2 sr 1 µm nm 16.3 nm at 63.2 W m 2 sr 1 µm nm 15.4 nm 200 at 58.1 W m 2 sr 1 µm nm 15.1 nm 200 at 48.7 W m 2 sr 1 µm nm 10.9 nm at 38.9 W m 2 sr 1 µm nm 38.1 nm 200 at 38.9 W m 2 sr 1 µm nm 33.7 nm 200 at 30.8 W m 2 sr 1 µm nm 21.1 nm at 27.5 W m 2 sr 1 µm nm 17.1 nm 200 at 22.6 W m 2 sr 1 µm High-resolution optical imager This instrument has the following main characteristics: (a) Spatial resolution in the range of less than one metre to several tens of metres; (b) Wavelengths in the VIS, NIR and SWIR bands (0.4 to 3 µm) with possible extension to MWIR and TIR; (c) Variable number of channels and bandwidths: (i) Single channel (panchromatic) with around 400 nm bandwidth (for example, nm);

16 16 VOLUME IV. SPACE-BASED OBSERVATIONS (ii) 3 10 multispectral channels with around 100 nm bandwidth; (iii) Continuous spectral range (hyperspectral); typically has 100 channels with around 10 nm bandwidth; (d) Imaging capability: continuous and contiguous sampling, covering a swath ranging from a few tens of kilometres to approximately 100 km, often addressable within a field of regard of several hundreds of kilometres; (e) Applicable in LEO. GEO is not excluded but not yet in use. High-resolution optical imagers may perform a number of missions, depending on the spectral bands, the number and bandwidth of channels, and steerable pointing capability. Those missions include: (a) Panchromatic imagers: surveillance, recognition, stereoscopy for digital elevation model, etc. Critical instrument features are the resolution and the steerable pointing capability; (b) Multispectral imagers: land observation for land use, land cover, ground water, vegetation classification, disaster monitoring, etc. Critical instrument features are the number of channels and the spectral coverage; (c) Hyperspectral imagers: land observation, especially for vegetation process study, carbon cycle, etc. Critical instrument features are the spectral resolution and the spectral coverage. Tables describe an example of a panchromatic imager (WV60), a multispectral imager (ETM+) and a hyperspectral imager (Hyperion). Table 3.7. Example of panchromatic high-resolution imager: WV60 on WorldView-1 WV60 World View 60 camera Satellite WorldView 1 Scanning technique Surveillance, recognition, stereoscopy for digital elevation model, etc. Panchromatic Resolution: 0.5 m Steering capability 60 cm telescope aperture Pushbroom detector array Swath: 17.6 km, addressable by tilting the satellite in a variety of operating modes Stereo capability both along-track and cross-orbits Global coverage in 6 months in daylight Global coverage in a few days (down to 3) by strategic pointing 0.50 m Mass: 380 kg Power: 250 W Data rate: 800 Mbps

17 CHAPTER 3. REMOTE-SENSING INSTRUMENTS 17 Table 3.8. Example of multispectral high-resolution imager: ETM+ on Landsat-7 ETM+ Enhanced Thematic Mapper + Satellite Scanning technique Landsat-7 Land observation for land use, land cover, ground water, vegetation classification, disaster monitoring, etc. 8 channels: 1 panchromatic, 6 VIS, NIR and SWIR, 1 TIR Resolution: 15 m, 30 m and 60 m Whiskbroom pixel/line (narrowband) pixel/line (panchromatic) pixel/line (TIR) Swath: 185 km Global coverage in 16 days in daylight 30 m (6 narrowband channels), 15 m (panchromatic), 60 m (TIR) Mass: 441 kg Power: 590 W Data rate: 150 Mbps Central wavelength Spectral interval SNR at specified input spectral radiance or NEΔT Low signal High signal Panchromatic µm 14 at 22.9 W m 2 sr 1 µm 1 80 at W m 2 sr 1 µm µm µm 36 at 40 W m 2 sr 1 µm at 190 W m 2 sr 1 µm µm µm 37 at 30 W m 2 sr 1 µm at W m 2 sr 1 µm µm µm 24 at 21.7 W m 2 sr 1 µm at W m 2 sr 1 µm µm µm 33 at 13.6 W m 2 sr 1 µm at W m 2 sr 1 µm µm µm 34 at 4.0 W m 2 sr 1 µm at 31.5 W m 2 sr 1 µm µm µm 27 at 1.7 W m 2 sr 1 µm at 11.1 W m 2 sr 1 µm µm µm 0.2 K at 300 K 0.2 K at 320 K Table 3.9. Example of hyperspectral high-resolution imager: Hyperion on NMP EO-1 Hyperion Satellite Scanning New Millennium Program Earth Observing 1 (NMP EO-1) Land observation, especially for vegetation process study, carbon cycle, etc. VIS/NIR/SWIR grating spectrometer with 220 channels (hyperspectral) in two groups, covering the ranges µm and µm respectively Channel bandwidth: 10 nm Resolution: 30 m Pushbroom

18 18 VOLUME IV. SPACE-BASED OBSERVATIONS technique 250 pixel/line Swath: 7.5 km Global coverage in one year in daylight 30 m Mass: 49 kg Power: 51 W Data rate: 105 Mbps SECTION: Ignore_book Chapter title in running head: CHAPTER 3. REMOTE-SENSING INSTRUMENTS Chapter_ID: 8_III_3_en Part title in running head: PART III. SPACE-BASED OBSERVATIONS Cross-nadir short-wave sounder An example is TROPOMI (TROPOspheric Monitoring Instrument) on the Sentinel-5 Precursor Satellite which has the following main characteristics: (a) The nadir viewing shortwave spectrometer of TROPOMI operates in the ultraviolet (UV), VIS, NIR and SWIR bands (see table below), the polarization sensitivity of TROPOMI is reduced to less than 0.5 % with a polarization scrambler; (b) Typical spectral resolutions of short-wave sounders are fractions of a nanometre (see Table 3.10); (c) Spatial resolution is in the order of 10 km - for TROPOMI it is 7 km x 7 km; (d) Horizontal sampling is not necessarily continuous and contiguous. However for TROPOMI it is contiguous thus providing imaging capability; (e) The scanning capabilities can be from nadir-only pointing to a swath of a few thousand kilometres. In the case of TROPOMI it has a swath of km; (f) Applicable both in LEO and GEO. Generally speaking cross-nadir SW sounders may be used in atmospheric chemistry for monitoring a number of species. Which species are detectable is determined by the spectral bands and necessitates sufficient spectral resolution. Good spatial resolution benefits sounding because it increases the number of clear-sky views. Contiguous sampling is a priori less critical however it has the great advantage of enabling imaging capabilities. Species that can be retrieved, depending on spectral coverage, include: (a) UV only: ozone profile; (b) UV and VIS: ozone profile and total column or gross profile of a small number of other species, such as BrO, NO 2, OClO, SO 2 and aerosol; (c) UV, VIS and NIR: ozone profile and total column or gross profile of several other species, such as BrO, ClO, H 2 O, HCHO, NO, NO 2, NO 3, O 2, O 4, OClO, SO 2 and aerosol; (d) UV, VIS, NIR and SWIR: ozone profile and total column or gross profile of many other species, such as BrO, CH 4, ClO, CO, CO 2, H 2 O, HCHO, N 2 O, NO, NO 2, NO 3, O 2, O 4, OClO, SO 2 and aerosol; (e) NIR and SWIR, possibly complemented by MWIR and TIR: total column or gross profile of selected species, such as CH 4, CO, CO 2, H 2 O and O 2.

19 CHAPTER 3. REMOTE-SENSING INSTRUMENTS 19 Tables 3.10 and 3.11 describe one example with in LEO its spectral coverage (TROPOMI) and another with its spectral coverage in GEO (UVN on Meteosat Third Generation). Table Example of cross-nadir SW sounder in LEO: TROPOMI instrument on Sentinel-5 Precursor (S-5 P) Spectral and Radiometric Performance Parameters for the TROPOMI Instrument TROPOMI Satellite The TROPOspheric Monitoring Instrument is a UV/VIS/NIR/SWIR nonscanning nadir viewing imaging spectrometer with a swath of about km and high spatial resolution (7 km x 7 km at nadir). TROPOMI flies in loose formation with with the SUOMI-NPP meteorological satellite 13:30h equator crossing time. Sentinel- 5 Precursor (S5-P) Atmospheric chemistry Mandatory products: O 3, SO 2, HCHO, NO 2, CO, CH 4, cloud fraction/pressure/optical depth, aerosol index and layer height Optional products: surface UV, aerosol optical depth, H 2 O, CHOCHO, BrO, OClO, HDO/ H 2 O ratio. Spectral range: UV/VIS/NIR/SWIR Imaging capability with 8 bands; grating spectrometer with polarization scrambler Scanning technique Push-broom scanner Swath: 2600 km; one scan line in 1 second Resolution Cross-track mode providing global coverage every day in daylight 7 km x 7 km at the SSP Mass: 200 kg Power: 170 W Data rate: 140 GBits per orbit

20 20 VOLUME IV. SPACE-BASED OBSERVATIONS Spectral Range Spectral resolution Spectral Sampling Required signal-tonoise Radiometric accuracy (%) nm 0.48 nm nm , nm 0.49 nm nm nm 0.54 nm 0.22 nm nm 0.54 nm 0.22 nm nm 0.38 nm 0.14 nm , nm 0.38 nm 0.14 nm , nm 0.25 nm 0.10 nm nm 0.25 nm 0.10 nm Note on spectral channels of TROPOMI: 1 : The minimum SNR is specified for a reference scene with a surface albedo of 2 % and the sun in zenith. 2 : The SNR for band 1 is specified for a ground pixel size of 21 x 28 km 2. 3 : The SNR for band 6 is specified for a ground pixel size of 7 x 7 km 2. 4 : The minimum SNR is specified for a reference scene with a surface albedo of 5 % and solar zenith angle of : Values refer to absolute radiometric accuracy of the measured Earth spectral reflectance. SOURCE:

21 CHAPTER 3. REMOTE-SENSING INSTRUMENTS 21 Table Example of cross-nadir scanning SW sounder in GEO: UVN on Meteosat Third Generation (MTG) UVN Satellites Scanning technique Ultraviolet, Visible and Near-infrared sounder Also known as Sentinel 4 MTG-S1 MTG-S2 Atmospheric chemistry Tracked species: BrO, ClO, H 2 O, HCHO, NO, NO 2, NO 3, O 2, O 3, O 4, OClO, SO 2 and aerosol Spectral range: UV/VIS/NIR Imaging capability: grating spectrometer covering 3 bands, channels Mechanical 3-axis stabilized satellite E W continuous S N stepping European area (lat. 30 N 65 N, long. 15 W 50 E) in 60 minutes (30 minutes is also possible) Resolution Defined at 45 N 0 : < 8 km in both N S and E W directions Mass: 150 kg Power: 100 W Data rate: 25 Mbps Spectral ranges No. of channels Spectral resolution SNR at specified input spectral radiance nm nm at W m 2 sr 1 µm nm nm at 140 W m 2 sr 1 µm nm nm at 60 W m 2 sr 1 µm Cross-nadir scanning infrared sounder These radiometers or spectrometers have the following main characteristics: (a) Wavelengths in the MWIR and TIR bands (3 15 µm) with a possible extension to FIR (up to 50 µm) and auxiliary channels in the VIS/NIR bands; (b) Spectral resolution in the order of 0.1 cm 1 (very high resolution), 0.5 cm 1 (hyperspectral) or 10 cm 1 (radiometer); (c) Spatial resolution in the order of 10 km; (d) Horizontal sampling not necessarily continuous or contiguous; (e) Scanning capability can be from nadir-only pointing to a swath of a few thousand kilometres; (f) Applicable both in LEO and GEO.